Sensitization via Healthy Skin Programs Th2 Responses in Individuals with Atopic Dermatitis

ORIGINAL ARTICLE

See related commentary on pg 2311

Sensitization via Healthy Skin Programs Th2 Responses in Individuals with Atopic Dermatitis Louise Newell1,3, Marta E. Polak1,3, Jay Perera1, Charlotte Owen1, Peter Boyd1, Christopher Pickard1, Peter H. Howarth1, Eugene Healy1, John W. Holloway1,2, Peter S. Friedmann1 and Michael R. Ardern-Jones1 Allergen-specific responses in atopic dermatitis (AD) are skewed toward a Th2 profile. However, individuals with AD have been shown to make effective virus-specific Th1 responses, raising the possibility that the skin itself contributes to driving the AD Th2 immunophenotype. Therefore, to explore the programming of immunological sensitization by the skin, we examined the outcome of sensitization through non-lesional skin of individuals with AD and healthy controls. Volunteers (controls, AD individuals with filaggrin gene (FLG) mutations (ADFM), and AD individuals without FLG mutations (ADWT)) were sensitized by cutaneous application of 2,4-dinitrochlorobenzene (DNCB), a small, highly lipophilic chemical sensitizer. At the doses tested, DNCB showed equal penetration into skin of all groups. Clinical reactions to DNCB were significantly reduced in AD. Although both controls and AD made systemic DNCB-specific Th1 responses, these were reduced in AD and associated with significantly Th2-skewed DNCB-specific T-cell responses. Th2 skewing was seen in both ADFM and ADWT, with no difference between these groups. After 3 months, DNCB-specific Th2 responses were persistent in individuals with AD, and Th1 responses persisted in controls. These data provide evidence that when antigen penetration is not limiting, AD skin has a specific propensity to Th2 programming, suggesting the existence of altered skin immune signaling that is AD-specific and independent of FLG status. Journal of Investigative Dermatology (2013) 133, 2372–2380; doi:10.1038/jid.2013.148; published online 9 May 2013

INTRODUCTION The immune responses of individuals with atopic dermatitis (AD) and atopic asthma are characterized by the generation of allergen-specific Th2 cells. Null mutations in the gene encoding filaggrin (FLG) occur in up to 50% of individuals with moderate-to-severe AD (Palmer et al., 2006; Barnes, 2010). This finding has redirected interest in understanding the role of the epidermis in the initiation of AD, which mirrors recent work suggesting the key role of defective bronchial epithelia in programming immune responses in atopic asthma (Xiao et al., 2011). Data demonstrating that skin barrier function is defective in AD skin (Jakasa et al., 2006; Proksch et al., 2006; O’Regan et al., 2010) and that impaired barrier function 1

Sir Henry Wellcome Laboratories, Clinical and Experimental Sciences, University of Southampton Faculty of Medicine, Southampton General Hospital, Southampton, UK and 2Human Genetics and Medical Genomics, University of Southampton Faculty of Medicine, Southampton General Hospital, Southampton, UK

3

These authors contributed equally to this work.

Correspondence: Michael R. Ardern-Jones, Sir Henry Wellcome Laboratories, Clinical and Experimental Sciences, University of Southampton Faculty of Medicine, Southampton General Hospital, Southampton SO16 6YD, UK. E-mail: [email protected] Abbreviations: ACD, allergic contact dermatitis; AD, atopic dermatitis; ADFM, atopic dermatitis with filaggrin mutations; ADWT, atopic dermatitis without filaggrin mutations; DNCB, 2,4-dinitrochlorobenzene; FLG, filaggrin gene; PBMC, peripheral blood mononuclear cell; PHA, phytohemagglutinin; SEB, Staphylococcal enterotoxin B; TSLP, thymic stromal lymphopoietin Received 11 June 2012; revised 6 February 2013; accepted 3 March 2013; accepted article preview online 25 March 2013; published online 9 May 2013

2372 Journal of Investigative Dermatology (2013), Volume 133

parallels disease severity (Nemoto-Hasebe et al., 2009) confirm that skin barrier function is important, but this does not fully explain the immunophenotype of AD. The Th1/Th2 paradigm (Mosmann et al., 1986; Umetsu et al., 1988; Romagnani, 2000) has been extensively explored in AD, and we, and others, have shown that, although allergenspecific Th1 cells can be detected in atopic and non-atopic individuals, allergen-specific Th2 cells are present in AD but not in non-atopic controls (Ardern-Jones et al., 2007). A mouse model has recently provided evidence that homozygous mutations in FLG are implicated in Th2 skewing (Fallon et al., 2009), and although genetic association between AD and the genes IL-4, IL-4R, and IL-13 has also been widely replicated, the precise reason for the bias toward a cutaneous Th2 response in AD remains obscure. Indeed, individuals with AD are not systemically immunocompromised, make appropriate vaccine-specific Th1-memory cells (Schneider et al., 2010), and have normal circulating Treg/Th17 frequency (Maggi et al., 2007; Koga et al., 2008). However, the fact that relative deficiency of Treg/Th17 cells has been identified in AD skin suggests that the primary immune defect in AD may be intrinsic to the epidermis (Verhagen et al., 2006; Guttman-Yassky et al., 2008). Several lines of evidence point to the skin as being centrally involved in programming the atopic immune responses, but up to now the definitive demonstration of this has been lacking. Thus, there is compelling evidence that the epidermal cytokine thymic stromal lymphopoietin (TSLP) can program dendritic cells via increased expression of OX40 to induce the & 2013 The Society for Investigative Dermatology

L Newell et al. Th2 Programming by AD Skin

generation of inflammatory Th2 cells characterized by the production of tumor necrosis factor-a in addition to the classical Th2 cytokines (Liu, 2006; Oyoshi et al., 2010). It has also been proposed that the Th2 allergen-specific responses observed in AD are a result of the defective barrier in the skin (O’Regan et al., 2008) or the chemical properties of the proteins themselves (Shakib et al., 2008). To address the question of whether primary epithelial regulation is critical in determining the quality of the allergic immune response, we used an experimental immunization system in humans delivered via the epidermis, and compared responses in AD and control subjects. Immunization or sensitization to environmental proteins would be unethical, and results may be biased by epidermal barrier function. Previous studies of the histological features of cutaneous responses to the model sensitizer 2,4-dinitrochlorobenzene (DNCB) in presensitized individuals document spongiosis and perivascular mononuclear infiltrate arising at 48 hours, which are the hallmarks of an eczematous response (Hartman et al., 1976). Therefore, using DNCB, which is not found in the environment, we could track the first exposure of an antigen through the skin and monitor the subsequent clinical and cellular immune priming of antigen-naive individuals to an allergen whose penetration was unaffected by the structural integrity of the skin barrier. RESULTS Penetration of DNCB is independent of FLG status

Basal/surface penetration

DNCB has found wide use in studying contact sensitization, as it is not present in the environment and can sensitize all individuals by cutaneous application (Friedmann et al., 1983). Covalent DNCB haptenation of host proteins is a prerequisite for the induction of contact hypersensitivity: the DNCBmodified proteins being processed and presented in both

class I and II major histocompatibility complex pathways to CD8 and CD4 T lymphocytes (Pickard et al., 2007). Consequently, tracking T-cell sensitization to DNCB gives a unique view of the initiation of immune responses generated by percutaneous antigen exposure in humans. DNCB is a lowmolecular-weight (202 Da) highly lipophilic molecule that penetrates the skin efficiently independent of skin structural barrier function (Gawkrodger et al., 1989; Falconer et al., 1992; Tsai et al., 2001). Equivalent penetration of DNCB at the sensitizing dose in AD and controls is critical to this study; therefore, we first set out to confirm this assumption. Using the established methodology developed by Franz et al. (2009), we tested the penetration of DNCB in human explant skin. In subsequent clinical studies, we used a sensitizing dose that has been shown to sensitize 100% of individuals (60 mg cm  2). However, to avoid missing subtle alterations in penetration, we used only 20 mg cm  2 in penetration assays, which would only sensitize approximately half the population (Friedmann et al., 1983). Following overnight penetration of DNCB through the surface of skin biopsies from healthy controls, individuals with AD without FLG mutations (ADWT), or individuals with AD with FLG mutations (ADFM), anti-DNCB fluorescence staining of histological sections was used to assess penetration of the antigen (Figure 1a–f). Although we cannot extrapolate these data to lower doses of DNCB, as expected, at one-third of the dose used in subsequent clinical experiments, equivalent strong penetration to the basement membrane was identified histologically between controls, and individuals with ADWT, and individuals with ADFM, and the relative fluorescence of anti-DNCB at the outermost viable epidermal level (base of stratum corneum) compared with the basal keratinocyte layer showed no significant difference between controls, individuals with ADWT, and individuals with ADFM (Figure 1h).

5

NS NS

NS

4 3 2 1 0 Controls

ADWT

ADFM

2

Figure 1. 2,4-Dinitrochlorobenzene (DNCB) penetration through the epidermis. (a, b) DNCB (20 mg cm ) was applied to the epidermis of skin biopsies from controls, (c, d) from individuals with atopic dermatitis without filaggrin mutation (ADWT), or (e, f) from filaggrin mutation heterozygote (ADFM) individuals. DNCB penetration was analyzed by immunohistochemistry with (a–f) anti-DNCB or (g) representative antibody control (green). Penetration of DNCB was assessed by measurement of the relative fluorescence of anti-DNCB (12 independent readings) from the base of the stratum corneum vs. the basal layer, as indicated by white dotted lines, (h) using image analysis software. Error bars represent mean±SEM. NS, no statistical significance. Scale bar ¼ 10 mm.

www.jidonline.org 2373

L Newell et al. Th2 Programming by AD Skin

Circulating lymphocytes in AD show a normal Th1/2 ratio

A Th2-polarized repertoire of circulating T cells in AD would be likely to modify immune priming toward Th2, and thus would complicate the investigation of the epithelial regulation of immune responses. Therefore, we first characterized the Th1/2 balance of circulating T cells in individuals with AD and controls in our population. First, we compared the production of IL-4 and IFN-g by ex vivo peripheral blood mononuclear cells (PBMCs) from individuals with AD (n ¼ 18) and healthy controls (n ¼ 10), and confirmed that resting unstimulated cells showed no difference in basal production of IL-4 (P ¼ 0.37) or IFN-g (P ¼ 0.94) (data not shown). We then compared non-TCR mitogen–induced T-cell activation (phytohemagglutinin (PHA)) against a direct TCR signal delivered by Staphylococcal enterotoxin B (SEB). SEB crosslinks specific SEB-binding TCR-Vb chains to major histocompatibility complex II, thereby inducing T-cell activation in a subset of responding T cells. In vitro activation showed that there was no significant difference in Th1 responses between PBMCs from both AD and controls when stimulated by PHA or SEB: P ¼ 0.06 and P ¼ 0.1, respectively (Figure 2). Importantly, circulating T cells in both AD and controls showed a similar frequency of Th2-polarized cells with no statistically significant difference in frequency or ratio between the groups (Figure 2). PBMC responses to house dust mite showed the expected Th2 programming in AD, which was absent in controls (Figure 2e). Therefore, it seems likely that the Th2-polarized responses to environmental proteins characteristic of AD are contained within a very small fraction of circulating lymphocytes, and because of such low frequency they are unlikely in themselves to perpetuate Th2 programming.

a

DNCB-specific Th1 and Th2 responses

To demonstrate that circulating DNCB-specific T cells could be characterized in DNCB-sensitized individuals, we first confirmed by ELISpot and flow cytometric double cytokine intracellular staining that both Th1 (Figure 4a and c) and Th2 (Figure 4d and f) DNCB-specific T-cell lines could be expanded in culture from the PBMCs of sensitized individuals to high frequencies. As expected, IFN-g-producing DNCBspecific T cells were not able to simultaneously produce IL-4 (Figure 4c). Similarly, IL-4-specific T cells were unable to produce IFN-g (Figure 4f).

100

100 10

10 1

IFN-γ SEB

1,000 Log10 SFU per 100,000 cells

Log10 SFU per 100,000 cells

One month after sensitization, participants were rechallenged on a different non-eczematous site by epicutaneous application of a dose series of DNCB, as we have previously described (Pickard et al., 2009). In concordance with previous findings (Rees et al., 1990), individuals with AD showed significantly reduced contact hypersensitivity responses to DNCB compared with healthy controls. This was reflected both by reduced inflammatory edema (skinfold thickness) (Figure 3a, P ¼ 0.007) and erythema (Figure 3b, P ¼ 0.015). Calculation of the area under the dose–response curve in each of healthy controls, individuals with ADFM, and individuals with ADWT, demonstrated that, compared with healthy controls, reduced cutaneous reactivity (skinfold thickness and erythema) was evident in both individuals with ADWT and individuals with ADFM (Figure 3c and d). ADFM showed further reduced skinfold responses to DNCB challenge as compared with ADWT, but this difference did not reach statistical significance (P ¼ 0.44, Figure 3c).

b

IFN-γ PHA

1,000

AD associates with impaired allergic contact dermatitis (ACD) responses to DNCB

2 3 4 6 Log10 PHA (μg ml–1)

10 AD Control

1 0.001 0.01 0.1 1 Log10 SEB (μg ml–1)

e d IL-4 PHA

100 10 1 1

2

3

4

IL-4 SEB

1,000 Log10 SFU per 100,000 cells

Log10 SFU per 100,000 cells

1,000

6

10

Log10 PHA (μg ml–1)

IL-4 SFU per 106 cells

c

100 10 1 0.001 0.01

10

*

400 300 200 100 0

0.1

1

Log10 SEB (μg ml–1)

10

AD

Control HDM

Figure 2. Circulating lymphocytes in atopic dermatitis (AD) show a normal Th1/2 ratio. (a, b) Sixteen-hour IFN-g ELISpot and (c, d) IL-4 ELISpot assay of ex vivo peripheral blood mononuclear cell (PBMC) activation with either phytohemagglutinin (PHA) (a, c) or Staphylococcal enterotoxin B (SEB) (b, d) in control individuals (gray lines) or in individuals with AD (black lines). (e) Ex vivo PBMC IL-4 ELISpot stimulated with house dust mite (HDM) extract. AD, n ¼ 18; Control, n ¼ 10. *Po0.05. SFU, spot-forming units.

2374 Journal of Investigative Dermatology (2013), Volume 133

L Newell et al. Th2 Programming by AD Skin

b 2.0 Controls n =10 AD n =16

110

1.0

0.5

90 70 50

0.0

30 6.25 12.5 8.8 17.7 DNCB challenge dose (μg)

17.7 6.25 8.8 12.5 DNCB challenge dose (μg)

c

d

P =0.01

50 40

P =0.016

2,000 AUC erythema index

AUC skinfold thickness (mm)

Controls n =10 AD n =16

1.5 Erythema index

Change in skinfold thickness (mm)

a

NS

30 20 10

1,500

NS

1,000 500

0

0 Control

ADFM

Control

ADWT

ADFM

ADWT

IFN-γ SFU per 106 cells

Figure 3. Atopic dermatitis (AD) subjects have impaired allergic contact dermatitis responses to 2,4-dinitrochlorobenzene (DNCB). Clinical response to epicutaneous challenge with DNCB 1 month following sensitization through the skin in AD (black lines) and controls (gray lines). Change in (a) skinfold thickness and (b) erythema index. (c) Area under the curve (AUC) for change in skinfold thickness and (d) erythema index for DNCB-elicitation reactions in controls or in individuals with AD with (ADFM) or without filaggrin mutations (ADWT). NS, no statistical significance.

105

3,000 2,000 1,000

0.0

0.1

104

104

103

103

102

102

101

0.0

0 101

Line

105

3,000 2,000 1,000 0 Ex vivo

Line

IFN-γ

IL-4 SFU per 106 cells

Ex vivo

105

102

103

104

0.5

105 104

103

103

102

102 2.9 101

102

103

104

105

1.8 101

104

101

0.8

101

105

0.0

6.0

102

103

104

0.2

105

0.0

101

5.1 101

102

103

104

105

IL-4

Figure 4. 2,4-Dinitrochlorobenzene (DNCB)-specific Th1 and Th2 cells. (a, d) DNCB-specific responses detected in peripheral blood mononuclear cells (PBMCs) (ex vivo; gray bars) or after short-term culture with DNCB (line; black bars) by (a) IFN-g or (d) IL-4 ELISpot from a DNCB-sensitized atopic individual. (b, c, e, f) Flow cytometric analysis of IFN-g or IL-4 production by (c, f) activated DNCB-expanded T-cell lines or (b, e) unstimulated control PBMCs. (a, c) Lowfrequency Th1 and (d, f) Th2 cells can be detected ex vivo, but higher frequency responses are detected after short-term culture expansion. Population frequency (%) as indicated in appropriate quartiles. Representative examples are shown. SFU, spot-forming units.

Sensitization through the skin induces skewing toward antigen-specific Th2 responses in AD

To examine the cytokine polarization of antigen-specific T cells following cutaneous sensitization with DNCB, PBMCs

from patients and controls were tested ex vivo for DNCBspecific IFN-g and IL-4 production by ELISpot assay (Figure 5a and b). No IFN-g or IL-4 DNCB-specific responses were detected in non-DNCB-sensitized controls (Figure 5a and b). www.jidonline.org 2375

L Newell et al. Th2 Programming by AD Skin

50 25

AD FM

T

D -A

-s

AD W

si en

N on

tiz

ed

AD

D

75

0

0

N on

-A

IL-4 SFU per 106 cells

25

N on

N on

-s

en

si

tiz

ed

0

50

P =0.02

100

AD

20

75

D

40

100

-A

60

NS

N on

IFN-γ SFU per 106 cells

IL-4 SFU per 106 cells

P =0.001

P =0.002

80

DNCB sensitized

NS IFN-γ SFU per 106 cells

IL-4 SFU per 106 cells

60

DNCB sensitized

DNCB sensitized

50 40 30 20 10 0

NS

100 80 60 40 20 0

Control

AD

Control

AD

Figure 5. Sensitization through the skin in atopic dermatitis (AD) shows unique skewing toward antigen-specific Th2 responses. ELISpot assay of 2,4dinitrochlorobenzene (DNCB)-specific (a) IL-4 or (b) IFN-g production by peripheral blood mononuclear cells (PBMCs) from non-sensitized controls (Nonsensitized horizontal striped bars), DNCB-sensitized controls without AD (Non-AD, black bars) (Non-AD), or DNCB-sensitized individuals with AD (AD, gray bars) 4 weeks after cutaneous DNCB sensitization AD. (c) DNCB-specific IL-4 from DNCB-sensitized controls (Non-AD), individuals without AD (ADWT) or with filaggrin mutations (ADFM). AD, n ¼ 16 (Supplementary Table S1 online); controls, n ¼ 10. Error bars represent mean±SEM. Long-lived DNCB-specific Th2 memory cells in AD shown by ELISpot assay of DNCB-specific (d) IL-4 or (e) IFN-g production by PBMCs from controls or individuals with AD 3 months after rechallenge with DNCB (4 months after sensitization). AD, n ¼ 10; sensitized controls, n ¼ 10; non-sensitized controls, n ¼ 6. Error bars represent mean±SEM. NS, no statistical significance; SFU, spot-forming units.

Although both groups demonstrated Th1 DNCB-specific responses (Figure 5b), induction of DNCB-specific Th2 responses were largely confined to AD individuals (P ¼ 0.02) (Figure 5a). Antigen-specific Th2 responses were significantly more frequent in both ADWT and ADFM (P ¼ 0.02 and 0.001, respectively, Figure 5c). ADFM showed a greater frequency of Th2-polarized DNCB-specific T cells compared with ADWT; however, this did not reach statistical significance (P ¼ 0.59; Figure 5c). Long-lived DNCB-specific Th2 memory cells in AD

Examination of the cohort 3 months after the rechallenge (4 months post sensitization) showed that DNCB-specific Th2 memory responses persist in individuals with AD (Figure 5d). However, memory-Th1 responses are diminished in the AD group at this stage, which is in contrast to healthy controls who maintain Th1-memory responses to DNCB (Figure 5e), although these differences were not statistically significant. DISCUSSION The important role of FLG-mediated epithelial barrier dysfunction in mediating AD is assumed, but the same genetic defect is also present in some healthy individuals without AD (9% European population (Palmer et al., 2006)) and Ichthyosis Vulgaris patients without AD (Smith et al., 2006). Indeed, it has been established that epidermal barrier function is 2376 Journal of Investigative Dermatology (2013), Volume 133

defective in both eczematous and clinically normal skin of individuals with AD, and application of allergens (e.g., house dust mite proteins) to the skin can induce skin lesions (Darsow and Ring, 2000; Ring et al., 2001) that are indistinguishable clinically and microscopically from AD (Langeveld-Wildschut et al., 1995, 1996), thus supporting an important role for allergic sensitization in AD pathogenesis. Furthermore, allergen avoidance has been shown by some to be at least partially effective in the treatment of AD (Clark and Adinoff, 1989; Fukuda et al., 1991; Sanda et al., 1992; Tan et al., 1996; Friedmann, 1999). Dysregulation of adaptive immunity in AD is reflected by Th2-biased allergen-specific responses, which are critical to B-cell class switching and induction of allergenspecific IgE production. We have confirmed that circulating Der p 1–specific Th2 cells are only present in atopic individuals including those with AD, whereas Der p 1– specific Th1 cells are detectable in both atopic subjects and healthy controls (Ardern-Jones et al., 2007). This suggests that although allergen has been encountered by both groups in a setting that initiates the expansion of allergen-specific T cells, the immune response is differently skewed in individuals with atopy and AD. In the atopic march described in children, AD usually arises first, most frequently in the absence of IgE sensitization (Kay et al., 1994; Illi et al., 2004), and is a major risk factor for the development of severe asthma and food allergy (Sporik et al., 1990; Illi et al., 2006). The development

L Newell et al. Th2 Programming by AD Skin

of specific IgE to environmental allergens is associated with defective skin barrier function (Boralevi et al., 2008), with skinrelated sensitization contributing to the association between AD and childhood asthma, as well as its persistence into adulthood (Burgess et al., 2008). Understanding the molecular pathway that drives Th2 polarization and in turn induces allergen-specific IgE production in individuals with AD is critical to the further development of therapeutic approaches for AD and potentially for atopic disease in general. Some previous studies have shown deficient Th1 pathways in circulating antigen-presenting cells/monocytes (IL-12) in AD, but others have shown opposing results (Itazawa et al., 2003; Piancatelli et al., 2008). We therefore set out to explore the immunological outcome of the same antigenic stimulus in the skin of individuals with AD versus non-atopic controls. As a chemical sensitizer, DNCB has a number of disadvantages as a model allergen for AD, because it may not replicate the cutaneous response to protein allergens. However, DNCB reliably sensitizes all individuals, and, importantly, in contrast to proteins, skin penetration at the sensitizing dose used here is not modified by AD or FLG status (Figure 1), allowing us to test epidermal regulation of immune responses between AD and controls without being confounded by disease status. Furthermore, DNCB-haptenated proteins require intracellular processing for presentation at the major histocompatibility complex in the immunological synapse, and therefore DNCB sensitization reflects the classical model of T-cell antigen recognition (Pickard et al., 2007). Here, we showed that clinical responses to DNCB were significantly attenuated in AD (Figure 3), suggesting that individuals with AD show a significantly impaired immune response to this antigen. Furthermore, ADFM and ADWT showed similar impairments in cutaneous immunity to DNCB, suggesting that the immune dysregulation is not FLG dependent but is due to other AD-specific epidermal factors and may be protective against more inflammatory Th1 responses classically associated with ACD. These findings are in line with clinical observations reporting a lack of association between FLG mutations with a phenotype of patients with both AD and ACD (Carlsen et al., 2011). Nickel ACD has been reported to associate with the presence of FLG mutations (Thyssen et al., 2008), but others have only found a nonsignificant trend toward this finding (Carlsen et al., 2010). The lack of histidine-rich FLG proteins leading to increased penetration of the metal could explain a nickel-specific association with loss of FLG protein in FLG mutations, but there is no evidence that other contact allergens bind FLG (Carlsen et al., 2010). In addition, FLG mutations impair impermeability of the epidermal water barrier, and because nickel is water soluble it is probable that percutaneous absorption of nickel will be greater in individuals with FLG mutations. It has previously been shown that higher doses of DNCB correlate with increased sensitizing potential, and therefore it would be expected that skin penetration is highly efficient in all individuals at a dose that sensitizes 100% of the population, as used here (Friedmann et al., 1983). To enhance the detection of differences in the penetration studies, we showed that a dose of DNCB that would be expected to sensitize

approximately half the population efficiently penetrated to the basement membrane in individuals with AD and controls. Although these data do not exclude the possibility of differences in DNCB penetration at lower concentrations, we feel that the equal DNCB penetration at the tested dose allowed us to directly compare the immune responses to DNCB in AD patients and healthy individuals. Therefore, in this study, we could be sure that priming of the immune response was initiated by equal cutaneous exposure to antigen in both AD and controls. DNCB would then be carried and presented by cutaneous dendritic cells for T-cell priming in the draining lymph node. In this way, we hoped to determine the effect of the skin on induction of subsequent immune responses in AD and controls without significant effects manifested by differences in other immunological compartments. Although variation in lymph node function could contribute to the observed differences, it has been previously shown that vaccination in the arm (bypassing the skin, utilizing axillary lymph node drainage as in this study) is equally effective at inducing Th1 responses in AD and controls (Leung et al., 1988; Schneider et al., 2010). Furthermore, we showed that in AD the circulating lymphocyte compartment was not uniquely skewed toward Th2 (Figure 2), as compared with controls, and previous reports have shown that serum levels of IL-4 in AD are normal (Takahashi et al., 1992). Despite the evident reduced skin responsiveness to DNCB in AD, both ADWT and ADFM showed significantly greater DNCB-specific Th2 responses than controls (Figure 5). Taken together, it seems likely that skewing of DNCB responses toward Th2 in AD is due to skin-specific regulation of DC signals determining polarization of immune responses to antigens encountered in the skin. We speculate that the altered function of cutaneous dendritic cells may explain the induction of DNCB-specific Th2 cells in AD, which would result in a less inflammatory response on antigen challenge, accounting for the impaired clinical responses seen here. We have also found that in biopsies from positive-DNCB skin challenge sites, gene transcription of IL-10 is higher in individuals with AD than in controls (data not shown), which may suggest that coregulation of regulatory cytokines in the skin is important in manifesting a reduced inflammatory reaction. Interestingly, controls were better able to maintain their Th1-memory population than AD patients (Figure 5), which we further speculate may suggest that altered programming by keratinocyte-derived TSLP in AD skin results in preferential maintenance of memory Th2 populations at the expense of Th1 skin–sensitized memory cells (Wang et al., 2006). These data would be predicted to be relevant for ACD. Recent evidence shows that ACD in AD is identified at least as frequently as in controls (Gittler et al., 2012). Our data demonstrate that once sensitized, individuals with AD are likely to elicit weaker responses on cutaneous re-exposure. This may suggest that the sensitivity of patch testing may be lower in AD, and thus allergic sensitization may be higher than that characterized by standard testing. The altered innate immune responses of the epithelia and particularly the epidermis suggest mechanisms by which www.jidonline.org 2377

L Newell et al. Th2 Programming by AD Skin

adaptive immune responses activated via the trans-epidermal route can be programmed or directed toward Th2 responses. Although there is deficient production of key components of antimicrobial innate defenses, IFN-a in the bronchial epithelium and beta defensins and cathelicidins in the epidermis, there is augmented production of prostaglandin E2 (Bao et al., 2011; Kalinski, 2012), GMCSF (Willart et al., 2012), and TSLP (Soumelis et al., 2002). These are all able to condition T cells toward the Th2 phenotype by modifying their activation by dendritic cells. Most particularly, TSLP can program dendritic cells via increased expression of OX40 to induce the generation of inflammatory Th2 cells characterized by the production of tumor necrosis factor-a in addition to the classical Th2 cytokines (Liu, 2006; Oyoshi et al., 2010). We wished to test the hypothesis that the epidermis of individuals with AD does indeed program a Th2 response to antigens encountered via the trans-epidermal route. We took the approach of inducing a new immune response with a widely used experimental organic contact sensitizer, DNCB, not found in the environment. A major advantage of DNCB is that as it is not water soluble its absorption into human epidermis would not be affected by the defective waterpermeability barrier found in many atopic patients—which we confirm. The normal human response to this and most small molecular contact sensitizers is known to be mediated by Th1 and Tc1 T cells (Pickard et al., 2007). We have been able to show that, in complete contrast to the normal, nonatopic response, when individuals with AD are sensitized through normal skin with DNCB, they indeed generate a substantial Th2 response to this agent. This demonstrates that the skin in humans with AD contributes to programming the adaptive T-cell response toward the Th2 type, which to our knowledge is previously unreported. This finding should direct future research to focus on the role of epithelia of skin and airways in programming the adaptive immune responses that characterize the atopic state. The data presented here demonstrate that AD skin is immunologically dysfunctional and that this manifests as impaired cutaneous reactivity to chemical antigens and is associated with Th2 polarization of the adaptive immune response. TSLP, a master regulator of Th2 polarization, has been shown to be upregulated in lesional, but not in nonlesional, AD skin (Soumelis et al., 2002); therefore, as these data were from non-lesional AD, they suggest a different AD-specific Th2 pathway. AD skin–specific Th2 skewing is of relevance to the disease pathogenesis, and characterization of key epidermal signals in AD skin critical to Th2 polarization will offer therapeutic targets for the T-cell–mediated allergic component of the disease and potentially for preventing future atopic disease in these individuals. MATERIALS AND METHODS Subjects Informed, written consent was obtained as per approval by the local research ethics committee (NRES07/Q1704/46) in adherence with the Helsinki guidelines. Adult AD patients with mild to severe disease (mean objective SCORAD 33.5±11.7, Supplementary Table S1 online) were recruited through the Dermatology Centre, Southampton 2378 Journal of Investigative Dermatology (2013), Volume 133

University Hospital NHS Trust, for comparison with a cohort of healthy non-atopic controls. All AD patients fulfilled the diagnostic criteria for AD as defined by the UK. AD diagnostic criteria (Williams et al., 1994a; Williams et al., 1994b; Williams et al., 1994c). Objective SCORAD was measured as described previously (Schmitt et al., 2007).

Quantification of contact sensitization by DNCB Individuals with AD and controls were sensitized to DNCB (Sigma, Poole, UK) at a concentration of 60 mg cm  2 in acetone by epicutaneous patch application to the upper inner arm skin at a non-lesional site (free from eczema) according to our previous method (Pickard et al., 2007, 2009). In all AD volunteers, this site showed no evidence of active eczema, and the volunteers were not being treated with topical therapy. Four weeks later, the strength of sensitization was determined by measurement of responses to simultaneous patch challenge with doses of 6.25, 8.8, 12.5, and 17.7 mg of DNCB to the opposite upper inner arm (free from eczema). Clinical responses were quantified 48 hours later at each challenge site by measuring the change in skinfold thickness with Harpenden skinfold calipers and erythema index (Dia-Stron, Andover, UK) as described previously (Patil et al., 1998; Paramasivan et al., 2009). Responses at challenged sites were expressed as a net increase after subtraction of baseline readings taken from adjacent normal skin. Values at each challenge site were summed to give a single value approximating to the area under the curve.

T-cell culture PBMCs were separated from whole blood by centrifugation on Lymphoprep (Axis-Shield PoC, Oslo, Norway); they were then washed and resuspended in R10HS: RPMI medium (Invitrogen, Carlsbad, CA) supplemented with penicillin per 100 IU ml  1, streptomycin 100 mg ml  1, and heat-inactivated pooled human serum (Sigma). DNCB was added at a final concentration of 2 mM. The cells were incubated at 37 1C with 5% CO2, with IL-2 (100 IU ml  1) and IL-7 (1 mg ml  1) supplemented at day 7. At day 13, these cells were harvested, washed, and returned overnight to wells containing R10HS.

ELISpot assays Ex vivo PBMCs were tested at a density of 2.5  105 cells per well; for T-cell lines, input cell numbers were 5  104 per well. DNCB was added at a final concentration of 2 mM. For PHA and SEB experiments, the final concentrations are shown in the figure (Figure 2). The plates were incubated overnight at 37 1C in 5% CO2 and were developed with streptavidin–alkaline phosphatase (Mabtech AB, Cincinnati, OH) and the alkaline phosphatase conjugate substrate kit (Bio-Rad, Hercules, CA). All experiments included vehicle as a negative control, and phytohemagglutinin (PHA) was used as a positive control. DNCBspecific responses were calculated by subtracting the counts from control wells, and the results were expressed as spot-forming units per million cells plated. Positive responses were recorded as those responses greater than the mean of all the background samples plus three times the SD of the background population.

Intracellular cytokine staining Culture-expanded DNCB-specific lymphocytes were activated with SEB (0.1 mg ml  1) and anti-CD28 (1 mg ml  1), with GolgiPlug

L Newell et al. Th2 Programming by AD Skin

(BD Biosciences, Oxford, UK). The Cytofix/Cytoperm kit (BD Biosciences) was used according to the manufacturers’ instructions. Flow cytometric analysis with the FACS Aria flow cytometer (BD Biosciences) was undertaken following lymphocyte gating on Forward/Side scatter and exclusion of dead cells (AQUA, Invitrogen, Paisley, UK). Subsequent gating on CD3-FITC and CD4-PerCP (BD Biosciences) was based on appropriate negative controls to demonstrate IFN-g- and IL-4-positive cells (Invitrogen and BD Biosciences) for analysis with the FlowJo software (Tree Star, Ashland, OR).

DNCB skin penetration and staining of frozen sections Skin explants from mastectomy tissue (healthy controls), AD FLG wild-type, and AD FLG variant subjects were mounted in modified Franz diffusion chambers. Skin penetration studies were undertaken as described previously (Gawkrodger et al., 1989; Franz et al., 2009). Briefly, DNCB (20 mg cm  2) in vehicle (acetone) was applied to the epidermal surface at 33% of the clinically used sensitizing dose to increase the possibility of identifying small differences in penetration. Explants were incubated overnight at 37 1C in 5% CO2 and then snap-frozen in liquid nitrogen. Frozen skin sections were cut using a cryostat and stained with a polyclonal goat anti-dinitrophenyl antibody (Sigma-Aldrich, Gillingham, UK; 1:2,000) for 1 hour at room temperature; next, an FITC-labeled anti-goat secondary antibody was added (Sigma-Aldrich; 1:400). Staining was visualized with fluorescent or confocal microscopy, and images were analyzed with ImageJ (Abramoff et al., 2004). The investigator undertaking image analysis was blinded to the subject disease or FLG status.

FLG genotyping FLG mutation analysis was performed as described previously (see Supplementary Material online for further information; Sandilands et al., 2007; Enomoto et al., 2008).

Statistical analysis See Supplementary Material online. CONFLICT OF INTEREST The authors state no conflict of interest.

ACKNOWLEDGMENTS We are grateful to the patients and volunteers involved in this study, and thank C McGuire, J Underwood, J Ward, AP Williams, Y Gao, G Di Genova, A Francisco Garcia, and T Sanchez-Elsner for technical advice. This work was funded by grants from the British Skin Foundation, Mason Medical Trust, and the Asthma, Allergy and Inflammation Research Charity. SUPPLEMENTARY MATERIAL

Barnes KC (2010) An update on the genetics of atopic dermatitis: scratching the surface in 2009. J Allergy Clin Immunol 125:16–29 Boralevi F, Hubiche T, Leaute-Labreze C et al. (2008) Epicutaneous aeroallergen sensitization in atopic dermatitis infants - determining the role of epidermal barrier impairment. Allergy 63:205–10 Burgess JA, Dharmage SC, Byrnes GB et al. (2008) Childhood eczema and asthma incidence and persistence: a cohort study from childhood to middle age. J Allergy Clin Immunol 122:280–5 Carlsen BC, Johansen JD, Menne T et al. (2010) Filaggrin null mutations and association with contact allergy and allergic contact dermatitis: results from a tertiary dermatology clinic. Contact Dermatitis 63:89–95 Carlsen BC, Thyssen JP, Menne T et al. (2011) Association between filaggrin null mutations and concomitant atopic dermatitis and contact allergy. Clin Exp Dermatol 36:467–72 Clark RA, Adinoff AD (1989) Aeroallergen contact can exacerbate atopic dermatitis: patch tests as a diagnostic tool. J Am Acad Dermatol 21:863–9 Darsow U, Ring J (2000) Airborne and dietary allergens in atopic eczema: a comprehensive review of diagnostic tests. Clin Exp Dermatol 25:544–51 Enomoto H, Hirata K, Otsuka K et al. (2008) Filaggrin null mutations are associated with atopic dermatitis and elevated levels of IgE in the Japanese population: a family and case-control study. J Hum Genet 53:615–21 Falconer AE, Friedmann PS, Bird P et al. (1992) Abnormal immunoglobulin G subclass production in response to keyhole limpet haemocyanin in atopic patients. Clin Exp Immunol 89:495–9 Fallon PG, Sasaki T, Sandilands A et al. (2009) A homozygous frameshift mutation in the mouse Flg gene facilitates enhanced percutaneous allergen priming. Nat Genet 41:602–8 Franz TJ, Lehman PA, Raney SG (2009) Use of excised human skin to assess the bioequivalence of topical products. Skin Pharmacol Physiol 22: 276–86 Friedmann PS (1999) Dust mite avoidance in atopic dermatitis. Clin Exp Dermatol 24:433–7 Friedmann PS, Moss C, Shuster S et al. (1983) Quantitative relationships between sensitizing dose of DNCB and reactivity in normal subjects. Clin Exp Immunol 53:709–15 Fukuda H, Imayama S, Okada K (1991) [The mite-free room (MFR) for the management of atopic dermatitis: living in the MFR improved first the itch and then the dermatitis]. Arerugi 40:626–32 Gawkrodger DJ, Haftek M, Botham PA et al. (1989) The hapten in contact hypersensitivity to dinitrochlorobenzene: immunoelectron microscopic and immunofluorescent studies. Dermatologica 178:126–30 Gittler JK, Krueger JG, Guttman-Yassky E (2012) Atopic dermatitis results in intrinsic barrier and immune abnormalities: implications for contact dermatitis. J Allergy Clin Immunol 131:300–13 Guttman-Yassky E, Lowes MA, Fuentes-Duculan J et al. (2008) Low expression of the IL-23/Th17 pathway in atopic dermatitis compared to psoriasis. J Immunol 181:7420–7 Hartman A, Hoedemaeker PJ, Nater JP (1976) Histological aspects of DNCB sensitization and challenge tests. Br J Dermatol 94:407–16 Illi S, von Mutius E, Lau S et al. (2004) The natural course of atopic dermatitis from birth to age 7 years and the association with asthma. J Allergy Clin Immunol 113:925–31

Supplementary material is linked to the online version of the paper at http:// www.nature.com/jid

Illi S, von Mutius E, Lau S et al. (2006) Perennial allergen sensitisation early in life and chronic asthma in children: a birth cohort study. Lancet 368: 763–70

REFERENCES

Itazawa T, Adachi Y, Okabe Y et al. (2003) Developmental changes in interleukin-12-producing ability by monocytes and their relevance to allergic diseases. Clin Exp Allergy 33:525–30

Abramoff MD, Magalhaes PJ, Ram SJ (2004) Image processing with ImageJ. Biophotonics Int 11:36–42 Ardern-Jones MR, Black AP, Bateman EA et al. (2007) Bacterial superantigen facilitates epithelial presentation of allergen to T helper 2 cells. Proc Natl Acad Sci USA 104:5557–62 Bao YS, Zhang P, Xie RJ et al. (2011) The regulation of CD4 þ T cell immune responses toward Th2 cell development by prostaglandin E2. Int Immunopharmacol 11:1599–605

Jakasa I, de Jongh CM, Verberk MM et al. (2006) Percutaneous penetration of sodium lauryl sulphate is increased in uninvolved skin of patients with atopic dermatitis compared with control subjects. Br J Dermatol 155:104–9 Kalinski P (2012) Regulation of immune responses by prostaglandin E2. J Immunol 188:21–8 Kay J, Gawkrodger DJ, Mortimer MJ et al. (1994) The prevalence of childhood atopic eczema in a general population. J Am Acad Dermatol 30:35–9

www.jidonline.org 2379

L Newell et al. Th2 Programming by AD Skin

Koga C, Kabashima K, Shiraishi N et al. (2008) Possible pathogenic role of Th17 cells for atopic dermatitis. J Invest Dermatol 128:2625–30

Romagnani S (2000) T-cell subsets (Th1 versus Th2). Ann Allergy Asthma Immunol 85:9–18

Langeveld-Wildschut EG, Thepen T, Bihari IC et al. (1996) Evaluation of the atopy patch test and the cutaneous late-phase reaction as relevant models for the study of allergic inflammation in patients with atopic eczema. J Allergy Clin Immunol 98:1019–27

Sanda T, Yasue T, Oohashi M et al. (1992) Effectiveness of house dust-mite allergen avoidance through clean room therapy in patients with atopic dermatitis. J Allergy Clin Immunol 89:653–7

Langeveld-Wildschut EG, van Marion AM, Thepen T et al. (1995) Evaluation of variables influencing the outcome of the atopy patch test. J Allergy Clin Immunol 96:66–73 Leung DY, Ambrosino DM, Arbeit RD et al. (1988) Impaired antibody responses in the hyperimmunoglobulin E syndrome. J Allergy Clin Immunol 81:1082–7 Liu YJ (2006) Thymic stromal lymphopoietin: master switch for allergic inflammation. J Exp Med 203:269–73 Maggi L, Santarlasci V, Liotta F et al. (2007) Demonstration of circulating allergen-specific CD4( þ )CD25(high)Foxp3( þ ) T-regulatory cells in both nonatopic and atopic individuals. J Allergy Clin Immunol 120:429–36 Mosmann TR, Cherwinski H, Bond MW et al. (1986) Two types of murine helper T cell clone. I. Definition according to profiles of lymphokine activities and secreted proteins. J Immunol 136:2348–57 Nemoto-Hasebe I, Akiyama M, Nomura T et al. (2009) Clinical severity correlates with impaired barrier in filaggrin-related eczema. J Invest Dermatol 129:682–9 O’Regan GM, Kemperman PM, Sandilands A et al. (2010) Raman profiles of the stratum corneum define 3 filaggrin genotype-determined atopic dermatitis endophenotypes. J Allergy Clin Immunol 126:574–80 O’Regan GM, Sandilands A, McLean WH et al. (2008) Filaggrin in atopic dermatitis. J Allergy Clin Immunol 122:689–93 Oyoshi MK, Larson RP, Ziegler SF et al. (2010) Mechanical injury polarizes skin dendritic cells to elicit a T(H)2 response by inducing cutaneous thymic stromal lymphopoietin expression. J Allergy Clin Immunol 126:976–84. 984 e971-975 Palmer CN, Irvine AD, Terron-Kwiatkowski A et al. (2006) Common loss-offunction variants of the epidermal barrier protein filaggrin are a major predisposing factor for atopic dermatitis. Nat Genet 38:441–6 Paramasivan P, Lai C, Pickard C et al. (2009) Repeated low-dose skin exposure is an effective sensitizing stimulus, a factor to be taken into account in predicting sensitization risk. Br J Dermatol 162:594–7

Sandilands A, Terron-Kwiatkowski A, Hull PR et al. (2007) Comprehensive analysis of the gene encoding filaggrin uncovers prevalent and rare mutations in ichthyosis vulgaris and atopic eczema. Nat Genet 39:650–4 Schmitt J, Langan S, Williams HC (2007) What are the best outcome measurements for atopic eczema? A systematic review. J Allergy Clin Immunol 120:1389–98 Schneider L, Weinberg A, Boguniewicz M et al. (2010) Immune response to varicella vaccine in children with atopic dermatitis compared with nonatopic controls. J Allergy Clin Immunol 126:1306–7 Shakib F, Ghaemmaghami AM, Sewell HF (2008) The molecular basis of allergenicity. Trends Immunol 29:633–42 Smith FJ, Irvine AD, Terron-Kwiatkowski A et al. (2006) Loss-of-function mutations in the gene encoding filaggrin cause ichthyosis vulgaris. Nat Genet 38:337–42 Soumelis V, Reche PA, Kanzler H et al. (2002) Human epithelial cells trigger dendritic cell mediated allergic inflammation by producing TSLP. Nat Immunol 3:673–80 Sporik R, Holgate ST, Platts-Mills TA et al. (1990) Exposure to house-dust mite allergen (Der p I) and the development of asthma in childhood. A prospective study. N Engl J Med 323:502–7 Takahashi T, Sasaki Y, Hama K et al. (1992) Production of IL-4, IL-2, IFNgamma, and TNF-alpha by peripheral blood mononuclear cells of patients with atopic dermatitis. J Dermatol Sci 3:172–80 Tan BB, Weald D, Strickland I et al. (1996) Double-blind controlled trial of effect of housedust-mite allergen avoidance on atopic dermatitis. Lancet 347:15–8 Thyssen JP, Carlsen BC, Menne T (2008) Nickel sensitization, hand eczema, and loss-of-function mutations in the filaggrin gene. Dermatitis 19:303–7 Tsai JC, Sheu HM, Hung PL et al. (2001) Effect of barrier disruption by acetone treatment on the permeability of compounds with various lipophilicities: implications for the permeability of compromised skin. J Pharm Sci 90:1242–54 Umetsu DT, Jabara HH, DeKruyff RH et al. (1988) Functional heterogeneity among human inducer T cell clones. J Immunol 140:4211–6

Patil SM, Patrick E, Maibach HI (1998) Animal, human, and in vitro test methods for predicting skin irritation. In: Marzulli FN, Maibach HI eds Dermatotoxicology Methods. The Laboratory Worker’s vade mecum. CRC Press: Taylor & Francis, 89–113

Verhagen J, Akdis M, Traidl-Hoffmann C et al. (2006) Absence of T-regulatory cell expression and function in atopic dermatitis skin. J Allergy Clin Immunol 117:176–83

Piancatelli D, Bellotta L, Del Beato T et al. (2008) Total IL-12 levels are increased in paediatric atopic dermatitis: correlations with age and disease severity. Int J Immunopathol Pharmacol 21:359–65

Wang YH, Ito T, Homey B et al. (2006) Maintenance and polarization of human TH2 central memory T cells by thymic stromal lymphopoietinactivated dendritic cells. Immunity 24:827–38

Pickard C, Louafi F, McGuire C et al. (2009) The cutaneous biochemical redox barrier: a component of the innate immune defenses against sensitization by highly reactive environmental xenobiotics. J Immunol 183:7576–84

Willart MA, Deswarte K, Pouliot P et al. (2012) Interleukin-1alpha controls allergic sensitization to inhaled house dust mite via the epithelial release of GM-CSF and IL-33. J Exp Med 209:1505–17

Pickard C, Smith AM, Cooper H et al. (2007) Investigation of mechanisms underlying the T-cell response to the hapten 2,4-dinitrochlorobenzene. J Invest Dermatol 127:630–7

Williams HC, Burney PG, Hay RJ et al. (1994a) The U.K. Working Party’s Diagnostic Criteria for Atopic Dermatitis. I. Derivation of a minimum set of discriminators for atopic dermatitis. Br J Dermatol 131:383–96

Proksch E, Folster-Holst R, Jensen JM (2006) Skin barrier function, epidermal proliferation and differentiation in eczema. J Dermatol Sci 43:159–69

Williams HC, Burney PG, Pembroke AC et al. (1994b) The U.K. Working Party’s Diagnostic Criteria for Atopic Dermatitis. III. Independent hospital validation. Br J Dermatol 131:406–16

Rees J, Friedmann PS, Matthews JN (1990) Contact sensitivity to dinitrochlorobenzene is impaired in atopic subjects. Controversy revisited. Arch Dermatol 126:1173–5 Ring J, Darsow U, Behrendt H (2001) Role of aeroallergens in atopic eczema: proof of concept with the atopy patch test. J Am Acad Dermatol 45:S49–52

2380 Journal of Investigative Dermatology (2013), Volume 133

Williams HC, Burney PG, Strachan D et al. (1994c) The U.K. Working Party’s Diagnostic Criteria for Atopic Dermatitis. II. Observer variation of clinical diagnosis and signs of atopic dermatitis. Br J Dermatol 131:397–405 Xiao C, Puddicombe SM, Field S et al. (2011) Defective epithelial barrier function in asthma. J Allergy Clin Immunol 128:549–56